🔬 optics

Near-infrared lensless holographic microscopy on a visible sensor enables label-free high-throughput imaging in strong scattering

This paper presents Near-infrared Lensless Digital Holographic Microscopy (NIR-LDHM), a cost-effective platform utilizing a standard visible CMOS sensor to achieve robust, label-free, high-throughput imaging of internal structures in strongly scattering biological tissues by leveraging near-infrared wavelengths and optimized sample-sensor distances.

Original authors: Emilia Wdowiak, Piotr Arcab, Mikolaj Rogalski, Anna Chwastowicz, Pawel Matryba, Malgorzata Lenarcik, Julianna Winnik, Piotr Zdankowski, Maciej Trusiak

Published 2026-02-17

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Emilia Wdowiak, Piotr Arcab, Mikolaj Rogalski, Anna Chwastowicz, Pawel Matryba, Malgorzata Lenarcik, Julianna Winnik, Piotr Zdankowski, Maciej Trusiak

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are trying to take a clear photograph of a secret message written on a piece of paper, but the paper is covered in thick, swirling fog. In the world of light microscopy, this "fog" is called scattering. When light hits a thick piece of tissue (like a slice of a mouse brain or liver), it bounces around chaotically, turning a clear image into a blurry mess.

For decades, scientists have had a powerful tool called Lensless Digital Holographic Microscopy (LDHM). Think of this as a special camera that doesn't use a glass lens to focus light. Instead, it uses math to reconstruct an image from the "shadows" and "ripples" of light that pass through an object. It's cheap, compact, and can see a huge area at once.

The Problem:
This magic trick usually only works if the "fog" is very thin. If you try to look through a thick slice of tissue, the light gets scattered so much that the math fails, and the image disappears. Also, most of these cameras are designed to see visible light (like the colors of a rainbow), which gets scattered very easily by tissue.

The Solution: The "Infrared Flashlight"
The researchers in this paper came up with a clever fix: Switch to Near-Infrared (NIR) light.

Think of visible light as a bouncy ball. When you throw it into a crowd of people (the tissue cells), it bounces off everyone and loses its direction. Now, think of Near-Infrared light as a slippery, heavy bowling ball. Because it's larger (has a longer wavelength), it doesn't bounce off the small "people" in the crowd as easily. It glides through the crowd much more smoothly, reaching the other side with its message intact.

The Surprise Twist: The "Farther is Better" Rule
Usually, in photography, you want your subject to be close to the camera for the best focus. But the scientists discovered a weird, counter-intuitive rule for this specific type of "foggy" imaging: Moving the camera farther away actually makes the image clearer.

The Analogy: Imagine you are standing in a room full of people shouting (the noise/scattering). If you stand right next to the speaker (the sample), the shouting is so loud and chaotic you can't hear the specific words. But if you walk to the back of the room, the chaotic shouting fades away faster than the speaker's voice. Suddenly, the voice becomes clear again.
The Science: By moving the sensor a bit further away, the "noise" (scattered light) spreads out and gets weaker, while the "signal" (the useful light carrying the image) stays relatively strong. This creates a clearer picture, even though the image is dimmer.

The "Low-Battery" Superpower
Another cool discovery is that this system works even when the "battery" is low.

Standard cameras need a lot of light to see infrared well. But this camera is so sensitive to the pattern of light (holography) that it can build a clear picture even with very few photons (light particles). It's like being able to read a book in a dark room just by feeling the bumps on the page, rather than needing a bright lamp.

The Real-World Test
The team tested this on real mouse tissues (liver and brain) without using any chemicals to clear them up (which is usually a messy, time-consuming process).

Visible Light: Could only see the outline of the tissue, like a silhouette.
Infrared Light: Saw right through the tissue, revealing tiny blood vessels and cell structures deep inside, even in slices up to 250 micrometers thick (which is huge for this kind of imaging).

Why This Matters
This is a big deal because:

It's Cheap: You don't need expensive, specialized infrared cameras. You can use a standard, low-cost camera sensor found in many devices.
It's Fast: You don't need to chemically treat the tissue first. You can look at fresh, real samples immediately.
It's Simple: The setup is small and doesn't need complex lenses.

In a Nutshell:
The researchers figured out how to see through thick, foggy biological tissue by using a special "slippery" infrared light and a clever trick of moving the camera slightly further away. They proved you can do this with cheap, standard equipment, opening the door for faster, cheaper, and deeper look-inside our bodies without damaging the samples.

1. Problem Statement

Lensless Digital Holographic Microscopy (LDHM) is a powerful, low-cost imaging technique that reconstructs complex optical fields (amplitude and phase) without imaging optics, offering large fields of view (FOV) and extended depth of field. However, classical LDHM relies on the Gabor in-line holographic model, which assumes a weak-scattering regime (single scattering).

The Limitation: In optically thick, heterogeneous biological tissues (e.g., brain or liver slices), multiple scattering occurs. This violates the weak-scattering assumption, causing the unscattered reference wave to be depleted and the object wave to become incoherent. Consequently, standard LDHM using visible (VIS) light fails to resolve features in samples thicker than a few hundred micrometers.
Current Solutions: Existing methods to image thick tissues often require:
- Optical clearing: An invasive, time-consuming chemical process that alters the tissue.
- Specialized NIR optics: Expensive, complex, and alignment-sensitive systems.
- Numerical corrections: Computationally intensive algorithms that struggle with strong scattering.

The authors aim to overcome these limitations by extending LDHM into the Near-Infrared (NIR) spectrum using standard, low-cost hardware, enabling label-free imaging of thick, uncleared tissues.

2. Methodology

The study introduces NIR-LDHM, a system that operates up to the silicon cutoff (~1100 nm) using a standard board-level CMOS sensor designed for visible imaging.

Hardware Setup:
- Illumination: A quasi-point source generating spherical waves. Six wavelengths were tested: 480, 532, 632 nm (VIS) and 785, 978, 1100 nm (NIR). Sources included supercontinuum lasers with tunable filters and laser diodes.
- Detector: A standard monochrome CMOS sensor (Allied Vision Alvium 1800 U-2050m, 2.4 µm pixels). Crucially, this sensor has very low Quantum Efficiency (QE) at 1100 nm (~0.19%) but was used without modification.
- Configuration: In-line geometry with the sample placed directly before the sensor.
Scattering Simulation:
- Phantoms: UHT pasteurized cow milk (3.2% fat) was used as a tissue-mimicking scattering medium due to its lipid-droplet structure and forward-scattering behavior.
- Setup: Scattering layers (SL) of controlled thickness (up to ~1.4 mm) were placed between the sample and the sensor to simulate strong scattering.
- Targets: USAF resolution targets (amplitude) and quantitative phase targets were used to measure resolution limits.
Reconstruction:
- Algorithm: Multi-plane Gerchberg-Saxton (GS) iterative algorithm using five holograms recorded at different sample-sensor distances.
- Low-Photon-Budget (LPB) Testing: Experiments were conducted at low signal levels (5–15 gray levels) to test robustness despite the sensor's poor NIR sensitivity.
Biological Validation:
- Thick, uncleared mouse tissue slices (liver: 20–60 µm; brain: 50–250 µm) were imaged to verify performance in real biological environments.

3. Key Contributions

NIR Extension of LDHM: Demonstrated that LDHM can operate effectively in the NIR spectrum (up to 1100 nm) using a standard VIS-optimized CMOS sensor, bypassing the need for expensive NIR detectors.
Discovery of a Distance-Dependent Mechanism: Identified a novel phenomenon where increasing the sample-sensor distance improves reconstruction quality under strong scattering. This contradicts the traditional LDHM design principle of minimizing distance to maximize Numerical Aperture (NA).
Label-Free Deep Imaging: Achieved high-throughput, label-free amplitude and phase imaging of thick, uncleared biological tissues (up to ~250 µm in brain tissue) without optical clearing.
Low-Photon-Budget Robustness: Proved that robust reconstructions are possible even with extremely low detector QE (0.19%) by leveraging the inherent noise tolerance of holographic reconstruction.

4. Key Results

A. Wavelength vs. Scattering Performance

VIS Failure: At 480–632 nm, resolution degraded rapidly. Features became unresolvable at scattering layer thicknesses of 180–360 µm.
NIR Success: At 785–1100 nm, the system maintained resolvable features through scattering layers up to ~1.4 mm (specifically 1400 µm at 1100 nm).
Resolution Comparison: At 1100 nm, no measurable resolution degradation was observed for scattering layers up to 570 µm.

B. The "Distance Mechanism" (Sample-Sensor Separation)

Observation: Under strong scattering, increasing the distance from the sample to the sensor (from ~3 mm to ~12–14 mm) improved lateral resolution by approximately twofold (e.g., from ~11 µm to ~5.5 µm).
Physical Explanation:
- Signal vs. Noise: The unscattered reference wave and weakly scattered signal occupy a narrower angular cone than the diffusely scattered noise.
- Geometric Filtering: As distance increases, the geometric spreading causes the amplitude of the diffuse noise (which has a wide angular distribution) to decay faster than the signal.
- SNR Improvement: At an intermediate distance (~12 mm), the noise is sufficiently suppressed to reveal higher spatial frequencies that were previously overwhelmed by noise at shorter distances.
- Limit: Beyond ~15–45 mm, the signal itself becomes too weak or low-NA, causing resolution to degrade again.

C. Biological Imaging

Mouse Liver: NIR (1100 nm) revealed internal trabecular patterns and vascular pathways up to 60 µm thickness. VIS (532 nm) failed to show internal structure beyond gross outlines at this depth.
Mouse Brain: NIR imaging resolved internal laminar structures up to 250 µm thickness. VIS imaging suffered from severe contrast loss and noise at thicknesses >150 µm.
Phase Imaging: Despite the reduced phase contrast at longer wavelengths (phase shift $\Delta\phi \propto 1/\lambda$ ), NIR-LDHM successfully recovered quantitative phase information in thick tissues.

5. Significance and Impact

Accessibility: By utilizing standard, low-cost VIS CMOS sensors, this approach makes deep-tissue quantitative phase imaging accessible without the high cost of specialized SWIR/NIR cameras or complex optical clearing protocols.
Clinical Potential: The ability to image thick, uncleared tissue slices rapidly and label-free is highly relevant for point-of-care diagnostics and intraoperative pathology, where speed and sample preservation are critical.
Theoretical Shift: The finding that increasing propagation distance can enhance resolution in turbid media reframes a fundamental design parameter in lensless holography, suggesting new strategies for "scattering-adapted" acquisition.
Scalability: The system offers a large Field of View (centimeter-scale) and is compact, making it suitable for high-throughput screening of biological samples.

In conclusion, the paper establishes NIR-LDHM as a robust, scalable, and cost-effective solution for imaging through strongly scattering media, bridging the gap between computational imaging and practical biophotonics applications.